Dario Amodei told JPMorgan Chase CEO Jamie Dimon on Friday that compute is “the deepest business dilemma that the company faces”, and that Anthropic handles it by buying “on the low side” in advance and paying more on the spot market when demand runs ahead. The two spoke at a fireside chat that Yahoo Finance streamed live from about 17:03 UTC on 13 November, five days before Anthropic’s IPO is expected to price.
Update, 14 November 2026, 17:50 UTC: Amodei on markets “much larger than coding”. Asked by Dimon whether parts of AI are becoming a commodity, Amodei said: “every time we extend the frontier, it creates entirely new markets.” He said that in 2023 “everyone was focused on chatbots”, and that Anthropic bet that “coding is gonna be much bigger than chatbots”. He added: “Now, coding is starting to get more mature. I think there’s still a long way to go in coding.” And: “I think the excitement is gonna move to these new areas which are gonna be much larger than coding in the same way that coding was much larger than chatbots.” The areas he named were “scientific discovery, life sciences”, “physics and materials, robotics, cyber security”, which he called “even more difficult challenges than vanilla coding”. Quotes are from our transcript of the Yahoo Finance stream.
Update, 14 November 2026, 13:20 UTC: Amodei on cyber defence by “swarms of agents”. Later in the same conversation, Dimon said cyber “was one of our biggest risks before mythos, before agents” and that “persistent agents are completely attacking almost every perimeter you have out there”, and asked what companies should do. Amodei described four steps. First, finding flaws: “Our customers have found hundreds of thousands of vulnerabilities.” Second, patching them, which “is harder than finding the vulnerabilities”. Third, “writing code that’s more secure in the first place”. The fourth, which Amodei said “hasn’t really happened to a great scale yet”, is real-time defence: “an attack happens over hours or minutes or maybe even seconds. And so having swarms of agents that sort of defensively patrol what a company has, a company’s servers and are able to defend against attacks in real time, this is something that I think has a lot of value.” Looking “six or 12 months from now”, Amodei said: “we’re going to need swarms in order to do the defense.” Quotes are from our transcript of the Yahoo Finance stream.
Update, 13 November 2026, 20:10 UTC: The Financial Times reported a further remark from the same event with Jamie Dimon. “We’re not super fussed on when exactly, what month or what day, or what market conditions we go public into,” Amodei told bankers and partners at the JPMorgan event, according to the FT. The FT said the listing is expected less than a week from Friday.
Anthropic’s S-1/A puts the bill at $900.9 billion of committed spend, substantially all of it compute and infrastructure. About half, $449.3 billion, is expected after 2030, by our arithmetic, and $70.7 billion is due in the 12 months after 30 September, against $131.9 billion of cash and short-term investments.
Business Pill · GPUs Explained
A short explainer on GPUs, the chips behind most AI compute. It teaches the general idea and does not discuss Anthropic’s commitments.
The key insight: As we read it, Amodei described a supply strategy, not a forecast: commit to the low case years ahead and pay up for short-term compute only when demand beats it. The S-1/A shows both halves, $900.9 billion of committed spend, about half of it after 2030, and $6.5 billion of short-term compute this year that costs significantly more, which is what the insurance he described has cost so far.
Dimon raised the topic about 14 minutes into the stream, saying Anthropic had disclosed “$900 billion of forward commitments for compute and leases and data centers”. He added that he had read that about 80% of compute goes to new models and about 20% to inference, and asked how Anthropic would get what it needscost-efficiently.
Amodei answered at 15:05: “compute is the deepest business dilemma that the company faces”. He tied it to growth he cannot forecast: “In 2024 and 2025, in both of those years, the company’s revenue grew by roughly 10x”, and yet “we have to buy compute a year or two years ahead of time”.
He set out both ways to get it wrong: “If you buy too much, of course, you can financially overextend yourself. If you buy too little, then you can’t support inference for customers”. Buying too little also means “you can’t do the R&D”, he said, which produces the model six or 12 months later.
When Anthropic expects to incur its $900.9 billion of committed spend, substantially all compute and infrastructure, per its S-1/A of 10 November 2026. The split between the $70.7 billion due in the next 12 months, the rest before 2031 and the remainder is our arithmetic from the filing.
Buy Low, Then Pay Up on the Spot Market
His answer, at 17:21: “we actually tend to buy on the low side”. Anthropic buys “relatively little compute for relatively low cost ahead of time”, he said, and if demand beats expectations it can “buy additional compute on the spot market, which so far has always been available and which we believe will continue to be available”.
“That costs more, but we’re buying it in worlds where we’re doing particularly well”, he said, likening the approach to “buying an insurance policy”.
He named two buffers between buying too much and too little. One is margins: “as you can see in the S1, our gross margins are quite good, especially compared to other players in the space”. The other is cash: “we have an enormous amount of cash on the balance sheet”.
How the S-1/A Lines Up
Anthropic is in its IPO quiet period, and its post announcing the chat carried the legend that it has “filed a registration statement (including a prospectus) with the SEC”. We checked each figure in the conversation against the S-1/A filed on 10 November.
The $900 billion. The filing gives an aggregate committed spend of about $900.9 billion as of and after 30 September 2026, “substantially all of which is related to compute and infrastructure”. It expects about $451.6 billion before 31 December 2030 and substantially all of the rest between 2031 and 2045; $70.7 billion is due in the next 12 months.
The ‘roughly 10x’. The filing’s revenue was $41 million in 2023, $386 million in 2024 and $4,585 million in 2025, which is 9.4 times and then 11.9 times the year before, by our arithmetic. Revenue in the nine months to 30 September 2026 was $33,242 million.
The spot market. The filing describes this compute as short-term commitments: $6.5 billion of compute under short-term commitments in the nine months to 30 September 2026, which “currently costs significantly more than comparable compute procured under our multi-year commitments”. In the third quarter, $5.0 billion of it “reduced revenue per dollar of compute compared to the preceding quarter”.
The gross margins. The phrase gross margin appears 0 times in the S-1/A, by our search. The filing reports compute and infrastructure as an operating expense: $18,808 million in the nine months to 30 September 2026, or 56.6% of revenue, against 172.1% a year earlier, by our arithmetic.
The cash. At 30 September Anthropic had $95.6 billion of cash and cash equivalents and $36.3 billion of short-term investments, $131.9 billion in all. The $70.7 billion of commitments due in the next 12 months is about 54% of that, by our arithmetic, before any IPO proceeds.
Anthropic’s annual revenue from its S-1/A of 10 November 2026, against Amodei’s remark that revenue grew by roughly 10x in both 2024 and 2025. The multiples are our arithmetic.
Why Go Public Now
Dimon asked “why now?” at about 28:43. Amodei said the biggest reason “goes back to these compute bets”: between the low compute world and the high compute world, “the more cash we have on hand, the more buffer we have between those two things”.
He said the advice that resonated most when preparing the listing was “never let your company be worth a single dollar more than it’s really worth”, which he likened to Anthropic’s habit to “under promise and over deliver”. Pricing is expected on 18 November and trading on 19 November, per Anthropic’s IPO website filed with the SEC; see our Anthropic IPO map.
Update (13 Nov): what Amodei told Dimon about culture. Earlier in the same fireside chat, by our transcript of the Yahoo Finance stream, Amodei said that from day one he and “the other six co-founders” decided to focus on talent and culture, with “a culture interview that everyone goes through”. He said Anthropic has been “very disciplined about turning down people who are deeply technically talented” but do not fit, adding: “we don’t want people who are jerks we don’t want people who try to steal the show”. He claimed “the best retention rates in the industry”, said “All seven co-founders of the company are still present” and that “Of the original 100 employees to join, more than three quarters are still at” the company, and put his own time on talent and culture at “probably 40, 50%”. Anthropic’s amended IPO filing backs the co-founder line: “Talent attraction and retention start with our seven co-founders, all of whom have remained with the company.” The industry-best retention, the first-100 figure and his time split are his claims; the documents we read do not let us check them. The filing also says “our seven co-founders are distinctly equipped to be stewards of our mission” and that they “have each pledged to dedicate 80% of their personal equity in Anthropic to philanthropic causes”.
Dimon’s Side of the Table
Dimon spoke as a customer: “we’re one of your largest customers”, he said, adding “We have 250,000 employees that use it many times a week”. JPMorgan is also on the deal. In the S-1/A, J.P. Morgan is the third underwriter named on the cover, after Goldman Sachs and Morgan Stanley, which are the representatives.
Affiliates of J.P. Morgan Securities LLC are among the joint lead arrangers of Anthropic’s $15.0 billion revolving credit facility, set up in September 2026 in place of a facility that started at $2.5 billion in May 2025, six times larger, by our arithmetic. Affiliates of JPMorgan and four other underwriters hold Anthropic shares, each group less than 0.5% of fully diluted shares, the filing says.
The Structural Read
As we read the filing, the spot-market top-up Amodei described matches what the S-1/A calls compute under short-term commitments: $5.0 billion of it in the third quarter reduced revenue per dollar of compute against the quarter before, the filing says.
As we read the exchange, Amodei’s reason for listing now and his compute answer are the same answer: on his account, more cash on hand widens the buffer between the low compute world and the high compute world. The $70.7 billion due in the next 12 months is about 54% of the $131.9 billion Anthropic held at 30 September, by our arithmetic.
Dario Amodei, with Jamie Dimon, 13 November 2026
“So you can think of it a little bit like we’re buying an insurance policy”
Three Implications
FOR COMPUTE About $451.6 billion of the $900.9 billion is expected before 31 December 2030 and $70.7 billion in the next 12 months, per the S-1/A.
FOR MARGINS The S-1/A reports compute and infrastructure at 56.6% of revenue in the nine months to 30 September 2026, against 172.1% a year earlier, by our arithmetic.
FOR THE TIMELINE Pricing is expected on 18 November and trading on 19 November, per Anthropic’s IPO website filed with the SEC, five and six days after the fireside chat, by our arithmetic.
The framework records: “the capex race is no longer a corporate spending cycle. It is a sovereign-scale financing problem dressed in tech-company clothing”.
As we read it, Amodei’s answer to Dimon puts one company inside that frame: $900.9 billion of commitments running to 2045, and an IPO he said is mainly about more cash to cushion the compute bet. The framework predates the IPO and does not discuss it.
What Is Not Established
The quotes are from our own recording of Yahoo Finance’s stream, transcribed by machine and checked against a second, larger transcription model for the passages quoted; timestamps are minutes into that recording. The link in Anthropic’s own post returned an unavailable-video error to our checks before and during the event.
Amodei gave no new financial figures. The 80% and 20% split of compute was Dimon’s, from something he said he had read; we did not find such a split in the S-1/A passages we searched. We did not contact Anthropic or JPMorgan for this piece.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
Amodei called compute Anthropic’s “deepest business dilemma” and said it buys “on the low side”, then pays more for spot compute in good years. The S-1/A shows $900.9 billion committed, about half of it after 2030, $6.5 billion of costlier short-term compute this year, and revenue that grew 9.4 times and 11.9 times, by our arithmetic.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. The quotes are from our own recording of Yahoo Finance’s YouTube live stream of the Amodei and Dimon fireside chat on 13 November 2026, which began at about 17:03 UTC and ran about 37 minutes; we transcribed it with faster-whisper and re-checked the quoted passages with a larger model. Timestamps are minutes into that recording. The figures are from Anthropic’s S-1/A filed with the SEC on 10 November 2026; the pricing and trading dates are from Anthropic’s IPO website, filed with the SEC as a free writing prospectus.
The legend quoted is from Anthropic’s post on X announcing the stream. The multiples, ratios, splits and the gross-margin search are our arithmetic and our search of the filing. Anthropic and JPMorgan were not contacted for this piece.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
